Monitoring the foreign exchange rate benchmark fix

ABSTRACT

We develop a Manipulation Index (ManIx) that captures the potential manipulation intention of dealers during the World Markets/Reuters (WMR) benchmark (London Close) period at 4 pm London time through a unique algorithm and simulation. The application of this model (using a dataset with dealers’ identities) can identify banks that are prone to potential manipulative behavior. The results concerning the identified banks are validated by the regulatory investigations. Implementation of this algorithm allows regulators better direct their limited resources towards more targeted in-depth investigation.

KEYWORDS:

• Simulation
• regulation technology
• WMR fixing
• FX market manipulation
• market microstructure

JEL CLASSIFICATIONS:

• G31
• G18
• O24

Acknowledgements

The authors are grateful for the valuable comments and suggestions from the two anonymous referees. We thank participants at International Conference of the Financial Engineering and Banking Society 2017, Conference on The Econometrics of Financial Markets 2017 as well as Albert Menkveld, Björn Hagströmer, and Kevin Keasey.

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Hossein Jahanshahloo http://orcid.org/0000-0003-0786-2415

Notes

1 The economic importance of the WMR fix rate arises from its application in the valuation of equity portfolios and derivatives (Evans 2018).

2 On 12 June 2013, Bloomberg news reported that traders in five banks manipulated the fix rate affecting $3,600,000,000,000 assets “in funds including pension and savings accounts that track global indexes” (Vaughan, Finch, and Choudhury 2013). The breaking of this news triggered 20 investigations in 10 countries and the European Union (Bloomberg Visual Data 2015) resulting in $11,800,000,000 fines as of December 2015.

3 The Bank for International Settlement’s estimated share of automated spot trading in the FX market is 24.7% of $4.124 billion (BIS 2011). Rime and Schrimpf (2013, 40) report that “EBS estimates that around 30%–35% of volume on its trading platform is HFT-driven.”

4 See Section 2.1 for further discussion of the quote driven nature of the FX market.

5 Although the Electronic Broker System (EBS) is known as the main venue for the price discovery of the Euro and Yen (Cabrera, Wang, and Yang 2009), Thomson Reuters claims that it matches the venues that are providing liquidity in the EBS (Thomson Reuters 2014).

6 From February 2015, the fixing period has increased from 1 minute (60 s) to 5 min (300 s).

7 It is difficult to observe dealers’ real intentions. In practice, we use the observed final median as a measure of dealers’ manipulation targets. This design captures only successful manipulative attempts. In other words, our approach is a conservative estimate of the true manipulation in the market.

8 Such a bootstrap design also has the important benefit of controlling for a size effect as we discuss later in the methodology session.

9 Although we take great comfort that the banks identified via our methodology are close to the regulatory investigations’ findings, we recognize that there may be false positives in our test. Not all the banks that we have identified have been investigated by the regulators. Whether or not these are false positive will only be found if regulators initiate investigations and release their findings.

10 “In response to the rapidly changing financial landscape the FCA set up Project Innovate and so regulators are working to keep up with new business models entering the financial system” (Government Office for Science 2015).

11 See Section 2.1 for further details.

12 These two ways that a quote becomes a trade shows that every trade is not necessarily executed at the best price.

13 Explanatory memorandum to the Financial Services and Markets Act 2000 (regulated activities) (amendment) order 2015 no. 369, section 2, by HM treasury.

14 “WM performs tolerance checks at the time the data are sourced and again after the calculation of the benchmark is complete. This may result in some captured data being excluded from the fix calculation” (The World Market Company 2015). Though the mid-price Fix rate is the most common form of use, the Fix rate is published as bid and ask prices. As a result, in our methodology we also examine the bid and ask quotes separately to analyze the behavior of dealers.

15 It is possible to quote a price lower than the current median while higher than the last quote price. Such a quote would be a less effective manipulation than a down tick quote (a quote that has a quote price lower than the previous quote price).

16 The division of 1 by 0.1334 yields 7.5, which means 1 quote out of 7.5 quotes contribute to fix rate discovery.

17 Note that only dealers that have at least one significant ManIx are included in the list.

18 We would like to thank the anonymous referee for suggestion of adding panel B to Table 4.

19 We remove the 15:59 because 30 s of the fixing window falls in this minute.

20 Although some examples of these banks’ manipulations were published by the FCA, there is no comprehensive information on the currencies that they manipulated.

21 Similarly, BoA and BNP Paribas are not in our database. These banks are more active in the EBS platform.

22 Since manipulative quotes in each given fix section has a common target median, banks having significant manipulative behaviors can be seen as having common intended manipulative goal.